Agentic Coding Skills

A unified repository for developing, testing, and deploying reusable skills for AI coding agents (Claude Code and OpenCode).

Overview

This repository provides a structured framework for building skills - composable, reusable capabilities that enhance AI coding agents. Skills combine instructions, helper scripts, templates, and examples to enable agents to perform complex, multi-step workflows consistently.

Why Skills?

• Consistency: Standardized workflows across different projects
• Reusability: Write once, use everywhere
• The Review Triad: A robust 3-tier architecture (Verify -> Test-Review -> Code-Review) for guaranteed quality
• Multi-Agent Support: Same skills work with Claude, Gemini, Codex, and OpenCode

Repository Structure

skills/
├── README.md                    # This file
├── WORKFLOW.md                  # Skill development workflow guide
├── skills/                      # Individual skill implementations
│   ├── worklog/                # Org-mode worklog generation
│   ├── update-spec-kit/        # Spec-kit ecosystem updates
│   └── template/               # Template for new skills
├── .specify/                    # Spec-kit framework (for developing this repo)
│   ├── templates/
│   ├── scripts/
│   └── memory/
└── .opencode/                   # OpenCode commands (for developing this repo)
    └── command/

Skills Included

Skill	Description	Status
code-review	The Critic: Multi-lens code review (bloat, smells, dead-code, redundancy, defense-in-depth).	Deployed
test-review	The Auditor: Audit test quality for flakiness, tautologies, and anti-patterns.	Deployed
verify-work	The Gatekeeper: Run project build/tests before claiming success. Enforces "The Iron Law".	Deployed
niri-window-capture	Invisibly capture screenshots of any window across workspaces using niri compositor.	Deployed
orch	Query multiple AI models for consensus decisions, second opinions, and devil's advocate analysis.	Deployed
screenshot-latest	Find and analyze the most recent screenshot without typing paths.	Deployed
worklog	Create comprehensive structured org-mode worklogs documenting work sessions.	Deployed
ai-tools-doctor	Check and sync AI tool versions against declared manifest.	Available
bd-issue-tracking	Track complex, multi-session work with dependency graphs using beads.	Available
doc-review	Lint markdown documentation for AI agent consumption using deterministic rules + LLM semantic checks.	Available
hq	Orchestrate multi-agent workflows using worker CLI and bd issue tracking.	Available
ops-review	Run multi-lens ops review on infrastructure files (Nix, shell, Docker, CI/CD).	Available
playwright-visit	Visit web pages using Playwright browser automation (screenshot, text, html, pdf).	Available
review-gate	Quality gate for cross-agent review enforcement. Blocks agent completion until approved.	Available
spec-review	Review spec-kit specifications using multi-model AI consensus before phase transitions.	Available
tufte-press	Generate Tufte-inspired study card JSON from conversation, build PDF, and print.	Available
update-opencode	Check for and apply OpenCode version updates in Nix-based dotfiles.	Available
update-spec-kit	Update spec-kit repository, CLI tool, and all project templates to latest.	Available
web-research	Conduct deep web research and produce structured reports.	Available
web-search	Search the web for information using Claude Code's subprocess capability.	Available
template	Template for creating new skills.	Development

Skill Structure

Each skill follows a standard structure:

skill-name/
├── SKILL.md                    # Primary skill definition
│   ├── Frontmatter metadata (name, description)
│   ├── When to Use section
│   ├── Process/Instructions section
│   └── Requirements section
├── README.md                   # User-facing documentation
├── scripts/                    # Helper scripts
│   ├── script1.sh
│   └── script2.sh
├── templates/                  # File templates
│   └── template-file.ext
└── examples/                   # Example outputs
    └── example-output.ext

SKILL.md Format

---
name: skill-name
description: One-line description of what this skill does
---

# Skill Name

Detailed description of the skill.

## When to Use

Explicit triggers and use cases for invoking this skill.

## Process

Step-by-step instructions for the agent to follow.

## Requirements

- Dependency 1
- Dependency 2

Quick Start

Using Skills

Claude Code: Skills in ~/.claude/skills/ are automatically available. The agent will invoke them when appropriate based on the description and When to Use sections.

OpenCode: Skills in ~/.config/opencode/skills/ are discovered by the opencode-skills plugin. Ensure the plugin is installed and configured.

Developing Skills

1. Create a new skill branch:

   git checkout -b feature/new-skill-name
   

2. Copy the template:

   cp -r skills/template skills/your-skill-name
   

3. Edit SKILL.md with your skill definition

4. Add scripts/templates as needed

5. Test locally by deploying to ~/.claude/skills/ or ~/.config/opencode/skills/

6. Document in README.md for users

See WORKFLOW.md for detailed development workflow.

Deployment

Claude Code Deployment

Manual deployment:

# Link skill to Claude Code skills directory
ln -s $(pwd)/skills/your-skill-name ~/.claude/skills/your-skill-name

Nix Home Manager deployment:

# In your home.nix or equivalent
home.file.".claude/skills/your-skill-name" = {
  source = /path/to/skills/skills/your-skill-name;
  recursive = true;
};

OpenCode Deployment

Manual deployment:

# Link skill to OpenCode skills directory
ln -s $(pwd)/skills/your-skill-name ~/.config/opencode/skills/your-skill-name

Nix Home Manager deployment:

# In your home.nix or equivalent
home.file.".config/opencode/skills/your-skill-name" = {
  source = /path/to/skills/skills/your-skill-name;
  recursive = true;
};

Enable opencode-skills plugin:

// ~/.config/opencode/config.json
{
  "plugin": ["opencode-skills"]
}

Design Principles

1. Agent-First Design

Skills are written for AI agents, not humans. Instructions should be:

• Explicit and unambiguous
• Procedural (step-by-step)
• Include decision criteria
• Reference specific files/commands

2. Self-Contained

Each skill should:

• Include all necessary scripts
• Document all dependencies
• Provide templates for generated content
• Include examples of expected output

3. Composable

Skills can invoke other skills, but should:

• Declare skill dependencies explicitly
• Handle missing dependencies gracefully
• Avoid circular dependencies

4. Testable

Skills should be testable by:

• Providing example inputs
• Defining expected outputs
• Including validation criteria
• Supporting dry-run modes where applicable

5. Version Controlled

All skill components should be:

• Tracked in git
• Documented with clear commit messages
• Tagged when significant changes occur
• Backward compatible when possible

Contributing

Adding a New Skill

1. Use the skill template as starting point
2. Follow the standard skill structure
3. Test with both Claude Code and OpenCode
4. Document in main README.md
5. Submit PR with comprehensive description

Modifying Existing Skills

1. Create feature branch
2. Update SKILL.md and associated files
3. Test changes with real-world usage
4. Update version/changelog if significant
5. Submit PR explaining changes and impact

Integration with Spec-Kit

This repository uses spec-kit for its own development workflow. The .specify/ and .opencode/ directories contain the spec-kit framework that enables structured feature development.

To develop a new skill using spec-kit:

# In OpenCode or Claude Code
/speckit.specify "Add a new skill for [description]"

This will guide you through the structured development process.

License

MIT

Related Projects

• Claude Code - Anthropic's official CLI for Claude
• OpenCode - Open source AI coding agent
• spec-kit - Specification-driven development framework
• opencode-skills - OpenCode skills plugin

Support

For issues, questions, or contributions, please open an issue in this repository.